John Carmack QuakeCon Keynote Speech Recap >

	2006/08/05 20:45:27: Posted by DM
	John Carmack, one of the founders of iD Software, gave his annual keynote address at QuakeCon yesterday. PlanetQuake4 has the recap posted for all of you who could not attend. One of the things Carmack mentioned is that even though iD's next game will be a 3D action game, they are "are branching out into different directions" with the project. This is certainly good to hear. The game will use an advanced version of the MegaTexture technology powering Quake Wars, and we all know how good that looks and sounds (click the news topic for media). >Link
	Labeled With  quake id software Comments [9]  | Rate this article: -- 1 2 3 4 5 6 7 8 9 10  | Avg. rating of 7.0

Sounds like Mr Carmack and ID is working on a 360 title and it will use a more advanced version of megatexturing *eye opens wide*

# id Software will be focusing on gameplay over graphics in their next unannounced title.

Holy game convention Batman, they're going to try and make a good game?

the use of multi processors and megatexturing are good news in terms of eye candy and performance....

but he gameplay is in first priority.

Nice, another bad port!!!!

If i ever see again shitty graphics and physics like those of Prey then im going to kill mr carmack because he has a big mouth!

Relax don't think Carmack has been sitting on his behind not learning up about the 360. I'm sure he's going to make the 360 really sing.

Summary taken from another forum. *credit to hasan*

hes bringing a brand new IP different from anything Id has ever done before to PC 360 and PS3 simultaneously

360 is the lead console development platform, PS3 is marginally more powerful than 360 but is also marginally more difficult to develop for.

PS3 is much more challenging to code for compared to 360 and PC...parallelism in PS3 is difficult compared to threading and multicore 360 and PCs....360 development is harder than PC due to 6 threads....optimization on parallelism and multicore systems is the key

Exploiting 360 and PS3 is still difficult. Audio coding is much better on consoles than PCs..



advantages PS3 and 360 have over PC is that you are in control of all your threads unlike PC where some if not many threads are dedicated to OS and other processes running on the system, PS3 and 360 being a closed design helps ....

IDs next game will be its first game to make the game faster and better if you have multicore systems like multicore PCs PS3 and 360....



Next Gen multicore systems will give Id a chance to split the logic in a way the game can sense the situation in the game and react to it accordingly...360 have 3 symmetrical processors where anyone of them can run the same code and you can set up pretty nice structures....but PS3 have asynchronous processors or assymmetric processors like the Cell, you need to think more about accelerator engines than processes...so in PS3 you do have 1 thread which goes through and does everything so you can run compression decompression code run well on PS3 but when you look how much faster can a game get you get small percentage numbers of increase compared to PC processors and thats AFTER you restructure things carefully around the Cell...

some ahmadll law is qouted saying even if you have 20% acceleration due to Cell like processors...it will not make the game that much faster..the hardware developers (aimed at IBM) just take thier own kernel and run it saying this and this wll be faster with multicore..multithreaded processors but that will not get the game running faster as it involves many different codes which might turn out to be slower...on next gen consoles you can make small applications whicih can run faster but no difference compared to PCs in game performance

he says what really makes the game go faster is better branch predictions and bigger caches...he think the light is near the end of the tunnel about how much faster the processors can really get



he says the future of processing is parallelism though as the orders of magnitude increase in processing speed will come to an end very soon and an example of parallism is the GPU

He says if Sony had 2 Cells one handling CPU processing and one GPU as they originally intended they would have failed this generation. he says they dodged the bullet on that one.

he says sampling is better than resolution which is too high currently"

This is what I like about Carmack he cuts through all the marketing bs and tells it like it is.



Correction his exact words were HOPEFULLY to 360 and ps3 simultaneously he didn't seem to fully confirm it.

From the sounds of it.. it doesn't sound like the extra spes will even contribute as much to the overall game's speed like people thought (even I thought it would help the overall speed of the game)

It seems it will more be accelerating specific aspects, but it wont necessarily take an entire game with all these special effects and amazing graphics going on and just make it faster as a result of numerous spe accelerated applications running. It can't be seen as processes on the ps3, but as accelerator engines. Basically I think this is 3 timese general purpose processing advantage Microsoft has been talking about surfacing as the 3 cores aren't seen as accelerator engines, but more as processes and thanks to all 3 being identical and capable of running the same code this is a processor that can really take an entire game and speed it up.

Especially with Carmack putting so much emphasis on parallelism as the way forward for the cpus on these consoles and that is far easier to obtain on the 360.

I don't think it would be out of line to say he's suggesting the 360 is the console with the most potential moving forward based on his emphasis on parallelism and the fact that all 3 360 cores are all capable of running the exact same code. He basically drove home in his keynote the reasons why he feels Microsoft made a better choice with the cpu inside the 360.

He's also saying don't trust IBM's benchmark results because they purposely setup the best case scenario for their performances analysis of the cpus and the impressive results they show aren't even reasons for why a game ends up not being fast so its ultimately useless. So having these impressive numbers here make no serious improvement to the game one way or the other. Video game development is hardly ever the best case scenario for a specific cpus strengths. They'll of course try their best to get the most impressive results, but he's saying they'll never achieve what those IBM setup results show.

He exact words in the quakecon keynote was "it winds up accelerating a few small applications, but has utterly negligible impact on the game performance"

As a matter of fact its best I don't assume what he meant, but thats what it sounded like he was saying.